Re: [PATCH v3 09/10] KVM: MMU: fix MTRR update

2015-05-14 Thread Paolo Bonzini
On 14/05/2015 02:16, Wanpeng Li wrote: > >SPTEs are kept around and cached. The "role" field is used as the hash > >key; if the role doesn't change, SPTEs are reused, so you have to zap > >the SPTEs explicitly. > > Btw, why the patch changelog mentioned that the root shadow page will be > reuse

Re: [PATCH v3 09/10] KVM: MMU: fix MTRR update

2015-05-13 Thread Wanpeng Li
On Wed, May 13, 2015 at 04:10:08PM +0200, Paolo Bonzini wrote: > > >On 13/05/2015 10:43, Wanpeng Li wrote: >> kvm_mmu_reset_context >> kvm_mmu_unload >> mmu_free_roots >> >> The original root shadow page will be freed in mmu_free_roots, where I >> miss? >> >> Another question maybe not rela

Re: [PATCH v3 09/10] KVM: MMU: fix MTRR update

2015-05-13 Thread Paolo Bonzini
On 13/05/2015 10:43, Wanpeng Li wrote: > kvm_mmu_reset_context > kvm_mmu_unload > mmu_free_roots > > The original root shadow page will be freed in mmu_free_roots, where I > miss? > > Another question maybe not related to this patch: > > If kvm_mmu_reset_context is just called to destroy

Re: [PATCH v3 09/10] KVM: MMU: fix MTRR update

2015-05-13 Thread Wanpeng Li
Hi Xiao, On Wed, May 13, 2015 at 02:42:27PM +0800, Xiao Guangrong wrote: >Currently, whenever guest MTRR registers are changed >kvm_mmu_reset_context is called to switch to the new root shadow page >table, however, it's useless since: >1) the cache type is not cached into shadow page's attribute so

[PATCH v3 09/10] KVM: MMU: fix MTRR update

2015-05-12 Thread Xiao Guangrong
Currently, whenever guest MTRR registers are changed kvm_mmu_reset_context is called to switch to the new root shadow page table, however, it's useless since: 1) the cache type is not cached into shadow page's attribute so that the original root shadow page will be reused 2) the cache type is s